Wow was FCP wrong

GooperMC · #1 09-23-2005, 04:35 PM

I really started to think about that hand that 2+2 and FCP were disagreeing (A35J on a 2h 4h Jc flop). After all that talk about putting the villain on a range of hands I decided to do exactly that.

I created a spread sheet that has a column for
- "all" the hands that the villain could have
- how far / ahead hero was to those hands
- probability that the villain had that hand

Then I started to play around and it was hard for me to make pushing all not +EV. Play around with this
spreadsheet and try to find some reasonable probabilities of hands that makes getting all in not profitable.

Wintermute · #2 09-23-2005, 04:49 PM

I took a look at that sheet, and my head exploded.

But in all seriousness, I do believe their way of thinking on that hand is not optimal... however, this whole discussion did bring up an interesting topic that I hasn't really been discussed at length here, which is how you should adjust your "range of hands weighting" based on the action of your opponent.

Mendacious just posted in another thread that he believes humans naturally overestimate the % of times they're crushed by their opponent, whereas a computer that does not take into account the implications of a big bet or raise from the opponent will "get" the % closer to correct. The computer's approach is obviously not optimal, but do others agree that the computer's approach is more optimal than the average, overly-scared human's?

In my mind, the business of assigning weights and breadth to this range of hands is probably what differentiates the good from the best of winning players. Unfortunately, it's probably a mainly intuitive, experiential skill. But I hope someone can prove me wrong and come up with a nice system.

TheWorstPlayer · #3 09-23-2005, 04:54 PM

Doesn't seem to me like a 'system' is possible. Which is why a bot will never be as good as a human. The proper weighting must take into account previous actions of the opponents (have you ever seen him semi-bluff the turn before?) as well as OPPONENT'S likely view of third parties in the hand (does he know CO is a maniac?) as well as current speed of opponent's play (is he tilting? did he just lose a big pot? is it late at night?) and many other factors which cannot really be quantified. I agree 100% that this skill differentiates the good from the great.

Ribbo · #4 09-23-2005, 05:00 PM

I just shove and let the other players make the tough decisions... Up here in north west England we call it "Feck It Poker". Started when the Asian Taxi drivers at our local casino were constantly heard muttering "Feck it, all-in" in 6 card omaha with nothing but a king high flush draw (I joke not about that hand)

Wintermute · #5 09-23-2005, 05:08 PM

I think the nature of PLO8 is that it's relatively easy to get to the point where you're beating the game pretty well WITHOUT a large degree of talent for this sort of weighting of range of hands business, so long as there is a reasonable amount of truly incompetent, or at least substantially sub-optimal players around. I'd say that I personally have gotten to that point for the low/middle limit games. But to make the leap to being able to thoroughly beat the higher limit games, where presumably many fewer truly bad players exist, this ability may be what proves to separate low/mid-limit "champs" from their money. Since all the literature from reputed sources like Steve Badger & Ray Zee seems to advocate a more reserved, tight approach to the game, I'm wondering if I have overdone the intentional correction to Mendacious's human over-estimated fear-of-your-opponent-having-you-dominated. That is, maybe I just need to fold more, thereby lessening the amount that I search for every miniscule +EV situation, in exchange for avoiding a few huge -EV situations.

Wintermute · #6 09-23-2005, 05:10 PM

[ QUOTE ]
I just shove and let the other players make the tough decisions... Up here in north west England we call it "Feck It Poker". Started when the Asian Taxi drivers at our local casino were constantly heard muttering "Feck it, all-in" in 6 card omaha with nothing but a king high flush draw (I joke not about that hand)

[/ QUOTE ]
Yeah, that's a large part of my approach, as you know. So far, it hasn't worked out too well at the higher limits... although it works well for middle limits, perhaps even more optimally than the "right" way to play the game, I think it might not be viable for higher limits.

emptyshell · #7 09-23-2005, 05:13 PM

Why the hell would you want someone to come up with a system?

Wintermute · #8 09-23-2005, 05:29 PM

[ QUOTE ]
Why the hell would you want someone to come up with a system?

[/ QUOTE ]
If you have the worst of it with respect to a certain skill in a game, wouldn't you prefer to neutralize the impact of that skill in the short-term?

Of course, though, this wasn't really what I was trying to get at--"system" wasn't the right word. I just meant that I would like to improve in this department... of course I'd like for the skill to still mean something once I eventually develop it.

Chamonyx · #9 09-23-2005, 05:37 PM

Thanks for the SS, Goops.

Do I understand correctly that it shows pushing here has an EV of about 80c relative to a $24 starting pot?

I also did a quick variance calculation: about 15.5, so a sigma of just under 4......

Mendacious · #10 09-23-2005, 05:40 PM

My point was that a computer will calculate the cold odds more correctly, but the human player who has suspicion will recognize more often when the lower probability event has actually occurred. Virtually everyone in this forum in the QQ99 example intuitively knew that someone had AQ even though "statistically speaking" it was more unlikely.
So when you resort to range of possible hand analysis and use statistics to determine probabilities, I think you are surrendering one of your most powerful intuitive weapons.